You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

json_tokener.c 12 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457
  1. /*
  2. * $Id: json_tokener.c,v 1.18 2006/01/26 02:16:28 mclark Exp $
  3. *
  4. * Copyright (c) 2004, 2005 Metaparadigm Pte. Ltd.
  5. * Michael Clark <michael@metaparadigm.com>
  6. *
  7. * This library is free software; you can redistribute it and/or modify
  8. * it under the terms of the MIT license. See COPYING for details.
  9. *
  10. */
  11. #include "config.h"
  12. #include <stdio.h>
  13. #include <stdlib.h>
  14. #include <ctype.h>
  15. #include <string.h>
  16. #include "bits.h"
  17. #include "debug.h"
  18. #include "printbuf.h"
  19. #include "arraylist.h"
  20. #include "json_object.h"
  21. #include "json_tokener.h"
  22. static struct json_object* json_tokener_do_parse(struct json_tokener *this);
  23. struct json_object* json_tokener_parse(char * s)
  24. {
  25. struct json_tokener tok;
  26. struct json_object* obj;
  27. tok.source = s;
  28. tok.pos = 0;
  29. tok.pb = printbuf_new();
  30. obj = json_tokener_do_parse(&tok);
  31. printbuf_free(tok.pb);
  32. return obj;
  33. }
  34. #if !HAVE_STRNDUP
  35. /* CAW: compliant version of strndup() */
  36. char* strndup(const char* str, size_t n)
  37. {
  38. if(str) {
  39. size_t len = strlen(str);
  40. size_t nn = min(len,n);
  41. char* s = (char*)malloc(sizeof(char) * (nn + 1));
  42. if(s) {
  43. memcpy(s, str, nn);
  44. s[nn] = '\0';
  45. }
  46. return s;
  47. }
  48. return NULL;
  49. }
  50. #endif
  51. static struct json_object* json_tokener_do_parse(struct json_tokener *this)
  52. {
  53. enum json_tokener_state state, saved_state;
  54. enum json_tokener_error err = json_tokener_success;
  55. struct json_object *current = NULL, *obj;
  56. char *obj_field_name = NULL;
  57. char quote_char;
  58. int deemed_double, start_offset;
  59. char c;
  60. state = json_tokener_state_eatws;
  61. saved_state = json_tokener_state_start;
  62. do {
  63. c = this->source[this->pos];
  64. switch(state) {
  65. case json_tokener_state_eatws:
  66. if(isspace(c)) {
  67. this->pos++;
  68. } else if(c == '/') {
  69. state = json_tokener_state_comment_start;
  70. start_offset = this->pos++;
  71. } else {
  72. state = saved_state;
  73. }
  74. break;
  75. case json_tokener_state_start:
  76. switch(c) {
  77. case '{':
  78. state = json_tokener_state_eatws;
  79. saved_state = json_tokener_state_object;
  80. current = json_object_new_object();
  81. this->pos++;
  82. break;
  83. case '[':
  84. state = json_tokener_state_eatws;
  85. saved_state = json_tokener_state_array;
  86. current = json_object_new_array();
  87. this->pos++;
  88. break;
  89. case 'N':
  90. case 'n':
  91. state = json_tokener_state_null;
  92. start_offset = this->pos++;
  93. break;
  94. case '"':
  95. case '\'':
  96. quote_char = c;
  97. printbuf_reset(this->pb);
  98. state = json_tokener_state_string;
  99. start_offset = ++this->pos;
  100. break;
  101. case 'T':
  102. case 't':
  103. case 'F':
  104. case 'f':
  105. state = json_tokener_state_boolean;
  106. start_offset = this->pos++;
  107. break;
  108. #if defined(__GNUC__)
  109. case '0' ... '9':
  110. #else
  111. case '0':
  112. case '1':
  113. case '2':
  114. case '3':
  115. case '4':
  116. case '5':
  117. case '6':
  118. case '7':
  119. case '8':
  120. case '9':
  121. #endif
  122. case '-':
  123. deemed_double = 0;
  124. state = json_tokener_state_number;
  125. start_offset = this->pos++;
  126. break;
  127. default:
  128. err = json_tokener_error_parse_unexpected;
  129. goto out;
  130. }
  131. break;
  132. case json_tokener_state_finish:
  133. goto out;
  134. case json_tokener_state_null:
  135. if(strncasecmp("null", this->source + start_offset,
  136. this->pos - start_offset))
  137. return error_ptr(-json_tokener_error_parse_null);
  138. if(this->pos - start_offset == 4) {
  139. current = NULL;
  140. saved_state = json_tokener_state_finish;
  141. state = json_tokener_state_eatws;
  142. } else {
  143. this->pos++;
  144. }
  145. break;
  146. case json_tokener_state_comment_start:
  147. if(c == '*') {
  148. state = json_tokener_state_comment;
  149. } else if(c == '/') {
  150. state = json_tokener_state_comment_eol;
  151. } else {
  152. err = json_tokener_error_parse_comment;
  153. goto out;
  154. }
  155. this->pos++;
  156. break;
  157. case json_tokener_state_comment:
  158. if(c == '*') state = json_tokener_state_comment_end;
  159. this->pos++;
  160. break;
  161. case json_tokener_state_comment_eol:
  162. if(c == '\n') {
  163. if(mc_get_debug()) {
  164. char *tmp = strndup(this->source + start_offset,
  165. this->pos - start_offset);
  166. mc_debug("json_tokener_comment: %s\n", tmp);
  167. free(tmp);
  168. }
  169. state = json_tokener_state_eatws;
  170. }
  171. this->pos++;
  172. break;
  173. case json_tokener_state_comment_end:
  174. if(c == '/') {
  175. if(mc_get_debug()) {
  176. char *tmp = strndup(this->source + start_offset,
  177. this->pos - start_offset + 1);
  178. mc_debug("json_tokener_comment: %s\n", tmp);
  179. free(tmp);
  180. }
  181. state = json_tokener_state_eatws;
  182. } else {
  183. state = json_tokener_state_comment;
  184. }
  185. this->pos++;
  186. break;
  187. case json_tokener_state_string:
  188. if(c == quote_char) {
  189. printbuf_memappend(this->pb, this->source + start_offset,
  190. this->pos - start_offset);
  191. current = json_object_new_string(this->pb->buf);
  192. saved_state = json_tokener_state_finish;
  193. state = json_tokener_state_eatws;
  194. } else if(c == '\\') {
  195. saved_state = json_tokener_state_string;
  196. state = json_tokener_state_string_escape;
  197. }
  198. this->pos++;
  199. break;
  200. case json_tokener_state_string_escape:
  201. switch(c) {
  202. case '"':
  203. case '\\':
  204. printbuf_memappend(this->pb, this->source + start_offset,
  205. this->pos - start_offset - 1);
  206. start_offset = this->pos++;
  207. state = saved_state;
  208. break;
  209. case 'b':
  210. case 'n':
  211. case 'r':
  212. case 't':
  213. printbuf_memappend(this->pb, this->source + start_offset,
  214. this->pos - start_offset - 1);
  215. if(c == 'b') printbuf_memappend(this->pb, "\b", 1);
  216. else if(c == 'n') printbuf_memappend(this->pb, "\n", 1);
  217. else if(c == 'r') printbuf_memappend(this->pb, "\r", 1);
  218. else if(c == 't') printbuf_memappend(this->pb, "\t", 1);
  219. start_offset = ++this->pos;
  220. state = saved_state;
  221. break;
  222. case 'u':
  223. printbuf_memappend(this->pb, this->source + start_offset,
  224. this->pos - start_offset - 1);
  225. start_offset = ++this->pos;
  226. state = json_tokener_state_escape_unicode;
  227. break;
  228. default:
  229. err = json_tokener_error_parse_string;
  230. goto out;
  231. }
  232. break;
  233. case json_tokener_state_escape_unicode:
  234. if(strchr(json_hex_chars, c)) {
  235. this->pos++;
  236. if(this->pos - start_offset == 4) {
  237. unsigned char utf_out[3];
  238. unsigned int ucs_char =
  239. (hexdigit(*(this->source + start_offset)) << 12) +
  240. (hexdigit(*(this->source + start_offset + 1)) << 8) +
  241. (hexdigit(*(this->source + start_offset + 2)) << 4) +
  242. hexdigit(*(this->source + start_offset + 3));
  243. if (ucs_char < 0x80) {
  244. utf_out[0] = ucs_char;
  245. printbuf_memappend(this->pb, (char*)utf_out, 1);
  246. } else if (ucs_char < 0x800) {
  247. utf_out[0] = 0xc0 | (ucs_char >> 6);
  248. utf_out[1] = 0x80 | (ucs_char & 0x3f);
  249. printbuf_memappend(this->pb, (char*)utf_out, 2);
  250. } else {
  251. utf_out[0] = 0xe0 | (ucs_char >> 12);
  252. utf_out[1] = 0x80 | ((ucs_char >> 6) & 0x3f);
  253. utf_out[2] = 0x80 | (ucs_char & 0x3f);
  254. printbuf_memappend(this->pb, (char*)utf_out, 3);
  255. }
  256. start_offset = this->pos;
  257. state = saved_state;
  258. }
  259. } else {
  260. err = json_tokener_error_parse_string;
  261. goto out;
  262. }
  263. break;
  264. case json_tokener_state_boolean:
  265. if(strncasecmp("true", this->source + start_offset,
  266. this->pos - start_offset) == 0) {
  267. if(this->pos - start_offset == 4) {
  268. current = json_object_new_boolean(1);
  269. saved_state = json_tokener_state_finish;
  270. state = json_tokener_state_eatws;
  271. } else {
  272. this->pos++;
  273. }
  274. } else if(strncasecmp("false", this->source + start_offset,
  275. this->pos - start_offset) == 0) {
  276. if(this->pos - start_offset == 5) {
  277. current = json_object_new_boolean(0);
  278. saved_state = json_tokener_state_finish;
  279. state = json_tokener_state_eatws;
  280. } else {
  281. this->pos++;
  282. }
  283. } else {
  284. err = json_tokener_error_parse_boolean;
  285. goto out;
  286. }
  287. break;
  288. case json_tokener_state_number:
  289. if(!c || !strchr(json_number_chars, c)) {
  290. int numi;
  291. double numd;
  292. char *tmp = strndup(this->source + start_offset,
  293. this->pos - start_offset);
  294. if(!deemed_double && sscanf(tmp, "%d", &numi) == 1) {
  295. current = json_object_new_int(numi);
  296. } else if(deemed_double && sscanf(tmp, "%lf", &numd) == 1) {
  297. current = json_object_new_double(numd);
  298. } else {
  299. free(tmp);
  300. err = json_tokener_error_parse_number;
  301. goto out;
  302. }
  303. free(tmp);
  304. saved_state = json_tokener_state_finish;
  305. state = json_tokener_state_eatws;
  306. } else {
  307. if(c == '.' || c == 'e') deemed_double = 1;
  308. this->pos++;
  309. }
  310. break;
  311. case json_tokener_state_array:
  312. if(c == ']') {
  313. this->pos++;
  314. saved_state = json_tokener_state_finish;
  315. state = json_tokener_state_eatws;
  316. } else {
  317. obj = json_tokener_do_parse(this);
  318. if(is_error(obj)) {
  319. err = -(enum json_tokener_error)obj;
  320. goto out;
  321. }
  322. json_object_array_add(current, obj);
  323. saved_state = json_tokener_state_array_sep;
  324. state = json_tokener_state_eatws;
  325. }
  326. break;
  327. case json_tokener_state_array_sep:
  328. if(c == ']') {
  329. this->pos++;
  330. saved_state = json_tokener_state_finish;
  331. state = json_tokener_state_eatws;
  332. } else if(c == ',') {
  333. this->pos++;
  334. saved_state = json_tokener_state_array;
  335. state = json_tokener_state_eatws;
  336. } else {
  337. json_object_put(current);
  338. return error_ptr(-json_tokener_error_parse_array);
  339. }
  340. break;
  341. case json_tokener_state_object:
  342. state = json_tokener_state_object_field_start;
  343. start_offset = this->pos;
  344. break;
  345. case json_tokener_state_object_field_start:
  346. if(c == '}') {
  347. this->pos++;
  348. saved_state = json_tokener_state_finish;
  349. state = json_tokener_state_eatws;
  350. } else if (c == '"' || c == '\'') {
  351. quote_char = c;
  352. printbuf_reset(this->pb);
  353. state = json_tokener_state_object_field;
  354. start_offset = ++this->pos;
  355. } else {
  356. err = json_tokener_error_parse_object;
  357. goto out;
  358. }
  359. break;
  360. case json_tokener_state_object_field:
  361. if(c == quote_char) {
  362. printbuf_memappend(this->pb, this->source + start_offset,
  363. this->pos - start_offset);
  364. obj_field_name = strdup(this->pb->buf);
  365. saved_state = json_tokener_state_object_field_end;
  366. state = json_tokener_state_eatws;
  367. } else if(c == '\\') {
  368. saved_state = json_tokener_state_object_field;
  369. state = json_tokener_state_string_escape;
  370. }
  371. this->pos++;
  372. break;
  373. case json_tokener_state_object_field_end:
  374. if(c == ':') {
  375. this->pos++;
  376. saved_state = json_tokener_state_object_value;
  377. state = json_tokener_state_eatws;
  378. } else {
  379. return error_ptr(-json_tokener_error_parse_object);
  380. }
  381. break;
  382. case json_tokener_state_object_value:
  383. obj = json_tokener_do_parse(this);
  384. if(is_error(obj)) {
  385. err = -(enum json_tokener_error)obj;
  386. goto out;
  387. }
  388. json_object_object_add(current, obj_field_name, obj);
  389. free(obj_field_name);
  390. obj_field_name = NULL;
  391. saved_state = json_tokener_state_object_sep;
  392. state = json_tokener_state_eatws;
  393. break;
  394. case json_tokener_state_object_sep:
  395. if(c == '}') {
  396. this->pos++;
  397. saved_state = json_tokener_state_finish;
  398. state = json_tokener_state_eatws;
  399. } else if(c == ',') {
  400. this->pos++;
  401. saved_state = json_tokener_state_object;
  402. state = json_tokener_state_eatws;
  403. } else {
  404. err = json_tokener_error_parse_object;
  405. goto out;
  406. }
  407. break;
  408. }
  409. } while(c);
  410. if(state != json_tokener_state_finish &&
  411. saved_state != json_tokener_state_finish)
  412. err = json_tokener_error_parse_eof;
  413. out:
  414. free(obj_field_name);
  415. if(err == json_tokener_success) return current;
  416. mc_debug("json_tokener_do_parse: error=%d state=%d char=%c\n",
  417. err, state, c);
  418. json_object_put(current);
  419. return error_ptr(-err);
  420. }

No Description

Contributors (1)